basebackup_retry_times incompatible with "backup_options = concurrent_backup"

[jeffjanes]

If you use concurrent_backup and basebackup_retry_times is greater than 0 and the retry does get triggered (for example, because an rsync fails), then the backup is doomed. It will eventually fail with the (somewhat misleading--it was not trying to issue a start backup command) message:

Asking PostgreSQL server to finalize the backup. ERROR: Backup failed issuing start backup command. DETAILS: Connection lost, reconnection not allowed

However, before failing with that message, it retries the rsync and so could spend huge amounts of time, disk activity, and network throughput on a doomed task.

It should refuse to do a retry when it already knows that reconnection is not allowed.